The returned column names are from the source. So if you have lazy operations then the
colnames here does not reflects the results of those operations. To obtain the correct names try
names(collect(get_chunk(df, 1)))
colnames(x, ...)# S3 method for disk.frame
names(x, ...)
# S3 method for disk.frame
colnames(x, ...)
# S3 method for default
colnames(x, ...)
a disk.frame
not used